© 2006 Nature Publishing Group

The distillation and volatility of ionic liquidsMartyn J. Earle1, Jose M.S.S. Esperanca2, Manuela A. Gilea1, Jose N. Canongia Lopes2, Luıs P.N. Rebelo2,Joseph W. Magee3, Kenneth R. Seddon1 & Jason A. Widegren3

It is widely believed that a defining characteristic of ionic liquids(or low-temperature molten salts) is that they exert nomeasurablevapour pressure, and hence cannot be distilled1,2. Here we demon-strate that this is unfounded, and that many ionic liquids can bedistilled at low pressure without decomposition. Ionic liquidsrepresent matter solely composed of ions, and so are perceived asnon-volatile substances. During the last decade, interest in thefield of ionic liquids has burgeoned3, producing a wealth ofintellectual and technological challenges and opportunities forthe production of new chemical and extractive processes4–6, fuelcells and batteries7, and new composite materials8,9. Much of thispotential is underpinned by their presumed involatility. Thischaracteristic, however, can severely restrict the attainabilityof high purity levels for ionic liquids (when they containpoorly volatile components) in recycling schemes, as well asexcluding their use in gas-phase processes. We anticipate thatour demonstration that some selected families of commonly usedaprotic ionic liquids can be distilled at 200–300 8C and lowpressure, with concomitant recovery of significant amounts ofpure substance, will permit these currently excluded applicationsto be realized.The belief that ionic liquids are not volatile can probably be traced

back to a report on first-generation ionic liquids10 by Øye and co-workers11, that the Franklin acidic [C2mim]Cl 2 AlCl3 system (inwhich [C2mim]þ is 1-ethyl-3-methylimidazolium) does not exhibit ameasurable vapour pressure, even at raised temperatures, whereasthe analogous NaCl-AlCl3 system exhibited a significant vapourpressure of Al2Cl6 (refs 11, 12). However, for [C2mim]Cl 2 AlCl3systems with a greater than 2:1 excess of AlCl3, a detectable vapour

pressure of Al2Cl6 was observed above 191 8C. Again, this is not avapour of ions, and is generated by dissociating the ionic liquid13.With the discovery of second-generation and third-generation ionicliquids, and the explosion of newly discovered systems, this lack ofvolatility seems to have been assumed rather than tested.Of course, it is known that ionic liquids can be thermally decom-

posed, leading to transalkylation (alkyl scrambling) in the conden-sate. For example, [C2mim]Cl thermally decomposes when heatedto 190 8C in vacuum: the volatile decomposition products (1-methyl-imidazole, 1-ethylimidazole, chloromethane, chloroethane, etheneand hydrogen chloride) can be recondensed to create a new ternaryionic liquid system: [C1mim]Cl 2 [C2mim]Cl 2 [C2eim]Cl, inwhich [C1mim]þ is 1,3-dimethylimidazolium and [C2eim]þ is1,3-diethylimidazolium14. Similarly, ionic liquids with protonatedcations, [BH]X (for example, [Hmim]Cl), will dissociate on heatingto produce the molecular base, B, and the molecular acid, HX, whichcan be recondensed to regenerate [BH]X15,16. However, in neitherscenario did the ionic liquid exert a vapour pressure—the vapourconsisted of uncharged molecular components, not the ionic liquiditself.We show here that a range of pure, aprotic, ionic liquids can be

vaporized under vacuum at 200–300 8C and then recondensed atlower temperatures. Some selected families of ionic liquids show nosigns of degradation either in the distillate or the residue. We alsodemonstrate significant enrichment in distillations of mixtures ofionic liquids. Multiple-ion cluster transfer into the gas phase is mostprobably the underlying mechanism by which these compounds firstevaporate, thenmaintain their stability in the low-density phase, and,finally, recondense as a pure ionic liquid at lower temperatures17.Two types of apparatus were used for these experiments: a

Kugelrohr apparatus (Fig. 1) and a sublimation apparatus (seeSupplementary Information). Survey experiments were performedat 300 8C and 0.1 to 5 mbar in the Kugelrohr apparatus(1mbar ¼ 100 Pa). These temperatures and pressures are below thepredicted boiling points18 of the ionic liquids, but it was possible tovaporize the ionic liquids and then condense them on colder parts ofthe Kugelrohr apparatus. Typically, times for the distillation were 4to 6 h. A photograph of the apparatus is given in Fig. 1. In thisapparatus, there is a possibility of splashing as a result of bubblesbursting in the ionic liquid, and to prevent this from affecting the

LETTERS

Figure 1 | Labelled photograph of the Kugelrohr oven and distillationapparatus. The central glassware rapidly rotates.

Table 1 | Distillation rates at 300 8C and 0.1mbar

Second-generation ionic liquid Distillation rate (g h21)

[C2mim][NTf2] 0.120[C10mim][NTf2] 0.070[C16mim][NTf2] 0.024

Results obtained in the Kugelrohr apparatus.

1The QUILL Centre, The Queen’s University of Belfast, Stranmillis Road, Belfast BT9 5AG, UK. 2Instituto de Tecnologia Quımica e Biologica, Universidade Nova de Lisboa, Av.Republica, Apartado 127, 2780-901 Oeiras, Portugal. 3Physical and Chemical Properties Division, National Institute of Standards and Technology, 325 Broadway, Boulder,Colorado 80305, USA.

Vol 439|16 February 2006|doi:10.1038/nature04451

831

© 2006 Nature Publishing Group

results glass wool was placed between the oven flask and thecollection flask.To demonstrate the distillation of pure ionic liquids, 1.0 g samples

of [C2mim][NTf2] and [C10mim][NTf2] (in which [C10mim]þ is1-decyl-3-methylimidazolium, and [NTf2]

2 is bis{(trifluoromethyl)-sulphonyl}amide, also known as bistriflamide, [N(SO2CF3)2]

2) wereheated for five hours at 300 8C and 0.1mbar. This process was filmedand two time-lapse (speeded up 100 and 150 times) video files areavailable in the Supplementary Information to demonstrate theprocess in action. [C16mim][NTf2] (in which [C16mim]þ is1-hexadecyl-3-methylimidazolium) was treated similarly. The con-densate was collected and analysed by 1H, 19F and 13C nuclearmagnetic resonance (NMR) spectroscopy. These three ionic liquidsare stable to prolonged heating at the distillation temperature andcould be distilled, although the distillation rate varied according tothe chemical structure (Table 1). The NMR spectra of these threecompounds showed that no detectable decomposition had occurredduring the distillations.Distillations of a range of other ionic liquids were performed at

300 8C in the Kugelrohr apparatus (Table 2). Generally, the bistri-flamide ionic liquids distilled without significant decompo-sition, and only salts containing the larger tetraalkylammonium([NCw x y z]

þ), tetraalkylphosphonium ([PCw x y z]þ), or cholinium

([Cx y zN(C2H4OH)]þ) cations showed signs of decomposition. Forthe triflate ([OTf]2) salts, only [C2dbu][OTf] (in which [Cndbu]

þ is1-alkyl-1,8-diazabicyclo[5.4.0]undec-7-enium) distilled without sig-nificant decomposition. The vapour pressures for triflate saltsseemed lower than for similar bistriflamide salts, as shown by theirlow distillation rates (,0.01 g h21). Examples of tosylate ([OTs]2,[Me-4-C6H4SO3]

2) and hexafluorophosphate ([PF6]2) ionic liquids

distilled very slowly, with little decomposition (the [PF6]2 salt

must be free of acidic or basic impurities for a clean distillation).Ionic liquids with other anions (for example, halides, sulphates, orcarboxylates) decomposed on distillation. The principal mechanismof decomposition was by dealkylation or transalkylation of thecation14. This is aided by a nucleophilic anion. The triflate anionhas low nucleophilicity, and hence forms relatively thermally stableionic liquids. The bistriflamide anion has an extremely low nucleo-philicity, and hence forms very thermally stable ionic liquids; this alsoleads to a very low probability of transfer of a proton from the cationto the anion.To probe the effects of temperature and pressure on the distilla-

tion of ionic liquids, three experiments were performed with[C6mim][NTf2] in a small glass sublimation apparatus (Table 3).No precautions for splashing were needed in this apparatus because it

is mechanically still and bubble formation was not observed at thetemperatures and pressures of these experiments. For each experi-ment, [C6mim][NTf2] was heated from room temperature to 200 or250 8C, then held at the maximum temperature for 1–2 h. Thetemperature at which condensation was first observed on the coldfinger condenser (or on the walls of the apparatus) was noted. At theend of the experiment, the relative rate of distillation was determinedfrom the amount of [C6mim][NTf2] that had condensed on the coldfinger. Additionally, the purities of the distillate and of the undistilledresidue were checked by 1H and 19F NMR spectroscopy. Thedistillation at atmospheric pressure (in dry air) and 250 8C failed inthat the distillation was very slow, and much of what ‘distilled’consisted of decomposition products. On the other hand, thedistillation at 200 8C and #0.001mbar proceeded much faster,despite being 50 8C lower in temperature, and subsequentNMR analysis showed that the distillate was pure [C6mim][NTf2].The pressure–temperature effects on the distillations are inti-mately related to the vapour pressure of the ionic liquids. A manu-script reporting the determination of the vapour pressure of[C12mim][NTf2] by a direct static method is under preparation,and a recent paper19 describing the use of an indirect (Knudsen)method for vapour pressure determination of [C4mim][NTf2] hasalso appeared.Thus, reduced pressure allowed the distillation to occur more

rapidly, at a lower temperature, and without observable decompo-sition. The unrecognized importance of low pressure to the success ofsuch distillations may have contributed to the myth that ionic liquidscannot be distilled—for example, differential scanning calorimetry(DSC) and thermal gravimetric analysis (TGA) studies (whichregister the thermal decomposition of ionic liquids) have all beenreported at atmospheric pressures.Thus we have presented experimental data to demonstrate that

several ionic liquids can be vaporized and recondensed withoutsignificant decomposition. We now consider the mechanism ofmass transfer. Below, we rule out the three conceivable alternativemechanisms of mass transfer that could explain our results withoutinvolving the vaporization of the ionic liquids as ionic species.We have eliminated the possibility of physical transfer (either by

‘spraying’, ‘splashing’ or ‘creeping’) by investigating the distillation ofequimolar binary mixtures of ionic liquids (see Table 4). Theobserved difference in the compositions of the distillate and residuein such experiments cannot be explained by a physical transfermechanism. For all of these distillations, the composition of thedistillate is enriched in one component (or in two ions for the case inwhich all four ions in the mixture are different) compared to theresidue, which is the hallmark of separative distillation.It could also be proposed that the ionic liquids were not being

transferred into the gas phase as ionic species, but were beingvolatilized as neutral molecular species by a proton transfer mecha-nism (for example, the BASIL process20 involves protonating1-methylimidazole to an ionic liquid, [Hmim]Cl, which dissociateson heating). Examination of Table 2 reveals that ionic liquids derivedfrom tetraalkylammonium cations, N,N-pyrrolidinium cations, and1-alkyl-1,8-diazabicyclo[5.4.0]undec-7-enium cations can be dis-tilled without significant decomposition. For these cation types,there is no feasible mechanism to transfer a proton to the anion; inother words, the only possible mechanism for volatilization of these

Table 2 | Effects of distillation at 300 8C in the Kugelrohr apparatus

Ionic liquid Conditions Fractionof residue

decomposed (%)

Fractionof distillate

decomposed (%)

[C4mim][NTf2] 8 h, 6 mbar 2 2[C4dmim][NTf2] 4 h, 6 mbar 1 1[C8dbu][NTf2] 4 h, 6 mbar 3 3[NC2 2 2 6][NTf2] 4 h, 6 mbar 1 50[NC2 2 2 2][NTf2] 4 h, 6 mbar 1 2[C4mpyrr][NTf2] 4 h, 6 mbar 1 1[PC6 6 6 14][NTf2] 4 h, 6 mbar 1 65[PC4 4 4 16][NTf2] 4 h, 6 mbar 1 60[C6 1 1N(C2H4OH)][NTf2] 4 h, 6 mbar 15 85[C2mim][OTf] 4 h, 7 mbar 4 12[C6mim][OTf] 4 h, 6 mbar 1 50[C2dbu][OTf] 4 h, 8 mbar 1 1[C4mim][PF6] 4 h, 6 mbar 1 1[C4mim][FAP] 18 h, 8 mbar 99 50[P(s-C4)3C1][OTs] 4 h, 6 mbar 5 5[C2mim][OSO3C2H5] 5 h, 7 mbar 99 99[PC6 6 6 14][decanoate] 6 h, 7 mbar 98 95

[Cnmpyrr]þ is 1-alkyl-1-methylpyrrolidinium; [FAP]2 is trifluorotriperfluoroethylphosphate. Afraction of 1% represents the limit of our detection, and is thus an upper bound.

Table 3 | Pressure effects on the distillation of [C6mim][NTf2]

p (mbar) tmax (8C) t at whichdistillation was

first observed (8C)

Distillation rate(mg h21)

Purity of thedistillate (%)

830 250 ^ 5 250 ^ 5 ,1 ,500.07 200 ^ 5 170 ^ 5 3 .99#0.001 200 ^ 5 155 ^ 5 15 .99

Results obtained in the sublimation apparatus.

LETTERS NATURE|Vol 439|16 February 2006

832

© 2006 Nature Publishing Group

ionic liquids is as intact ions, either alone or aggregated. This isrepresented schematically in Fig. 2.In the case of tetraalkylphosphonium-based ionic liquids, the

majority decomposed before distillation could occur. However, inthe case of tris(sec-butyl)methylphosphonium tosylate, distillationwithout significant decomposition was observed, and again there isno possible mechanism which can be envisaged involving protontransfer. Finally, the possibility of dissociative alkyl transfer waseliminated, as there was no evidence for alkyl scrambling in anyexperiment (such as when [C2mim]Cl is distilled)14.In the cases of 1-alkyl-3-methylimidazolium salts, the majority

distilled without decomposition. However, here a proton transfermechanism must be considered, as the possibility of transferring the2-H proton from the cation to the anion, creating a carbene and afree acid, is feasible. Nonetheless, a study of the decomposition of[C2mim][NTf2] by pyrolysis mass spectrometry found no evidencefor the formation of HNTf2 at 300 8C (ref. 21). We further note thatwhenwe replace the 2-H proton on the cationwith a 2-methyl group,the new ionic liquid [C4dmim][NTf2] (in which [C4dmim]þ is1-butyl-2,3-dimethylimidazolium) is both volatile and less decom-posed than [C4mim][NTf2] (see Table 2). If a proton transfermechanism were dominant, the 2-methylated ionic liquid wouldhave had its volatility suppressed with a much greater level ofdecomposition. Indeed, examination of Table 4 reveals that ina mixture of tetraethylammonium bistriflamide and 1-ethyl-3-methylimidazolium triflate, a tetraethylammonium ionic liquid

vaporizes preferentially to an imidazolium ionic liquid, making aproton transfer mechanism appear highly improbable.Thus, if we examine all the reported cases, we can completely

discount direct physical transfer of liquids, and also eliminatethe possibility of dissociative mechanisms (either alkyl or protontransfer) for all ionic liquids except those based around the 1-alkyl-3-methylimidazolium cation, and even in that case proton transfermust be considered to be highly unlikely. For these latter ionicliquids, we believe that there is no significant contribution fromdissociation of the cation to form carbenes, but at this stage it cannotbe entirely eliminated.These distillations, accompanied by the recovery of substantial

amounts of commonly employed ionic liquids, corroborate recenttheoretical predictions18 that some selected families of ionic liquidscould boil at a temperature sufficiently low to avoid decomposition.These observations lay to rest a paralysing and invalid assumptionthat has dominated and restricted the field of ionic liquids since itsorigins, and should open up new ways for exploiting their properties.At the very least, a new method for purifying ionic liquids now exists,and we can imagine other applications (such as isolation of highlysoluble products by high-temperature crystallization). However,near ambient temperature the vapour pressure of ionic liquidsremains negligible, so we have, in effect, the best of both worlds.

Received 26 August; accepted 21 November 2005.

1. Seddon, K. R. Room-temperature ionic liquids—neoteric solvents for cleancatalysis. Kinet. Catal. 37, 693–-697 (1996).

2. Seddon, K. R. in Molten Salt Forum: Proceedings of 5th International Conference onMolten Salt Chemistry and Technology (ed. Wendt, H.) 53–-62 (Trans TechPublications, Zurich, 1998).

3. ISI Essential Science Indicators Special Topics: Ionic Liquids khttp://www.esi-topics.com/ionic-liquids/index.htmll (ISI, 2004).

4. Rogers, R. D. & Seddon, K. R. (eds) Ionic Liquids IIIB: Fundamentals, Progress,Challenges, and Opportunities—Transformations and Processes (AmericanChemical Society, Washington DC, 2005).

5. Rogers, R. D. & Seddon, K. R. (eds) Ionic Liquids IIIA: Fundamentals, Progress,Challenges, and Opportunities—Properties and Structure (American ChemicalSociety, Washington DC, 2005).

6. Atkins, M. P. et al. Ionic Liquids: A Map for Industrial Innovation: Q001 (January2004) (QUILL, Belfast, 2004).

7. Ohno, H. (ed.) Electrochemical Aspects of Ionic Liquids (Wiley-Interscience,Hoboken, 2005).

8. Rogers, R. D. & Seddon, K. R. (eds) Ionic Liquids: Industrial Applications for GreenChemistry (American Chemical Society, Washington DC, 2002).

9. Rogers, R. D., Seddon, K. R. & Volkov, S. (eds) Green Industrial Applications ofIonic Liquids (Kluwer, Dordrecht, 2002).

10. Wasserscheid, P. & Welton, T. (eds) Ionic Liquids in Synthesis (Wiley-VCH,Weinheim, 2003).

11. Dymek, C. J. Jr, Hussey, C. L., Wilkes, J. S. & Øye, H. A. in Joint (Sixth)International Symposium on Molten Salts (eds Mamantov, G. et al.) 93–-104 (TheElectrochemical Society, Pennington, New Jersey, 1987).

12. Øye, H. A. & Wahnsiedler, W. E. in The Fourth International Symposium onMolten Salts (eds Blander, M., Newman, D. S., Saboungi, M. L., Mamantov, G. &Johnson, K.) 58–-73 (The Electrochemical Society, Pennington, New Jersey,1984).

13. Øye, H. A., Jagtoyen, M., Oksefjell, T. & Wilkes, J. S. Vapour pressure andthermodynamics of the system 1-methyl-3-ethylimidazolium chloride-aluminium chloride. Mater. Sci. Forum 73–-75, 183–-190 (1991).

Table 4 | Distillation of approximately equimolar mixtures of ionic liquids

Ionic liquid A Ionic liquid B Composition ratios ofthe residue (A:B)

Composition ratios ofthe distillate (A:B)

Fraction of distillatedecomposed (%)

[C2mim][NTf2]†‡ [C16mim][NTf2] 24:76 76: 24 ,1[C2mim][NTf2]§ [C6mim][NTf2] 5 1:49 59:41 ,1[C4mpyrr][NTf2]‡ [NC2 2 2 2][NTf2] 47:53 53:47 ,1[C4mim][NTf2]§ [C4mim][PF6] 52:48 98:2 ,1[C2mim][OTf]‡ [NC2 2 2 2][NTf2] * * ,1

†A mass balance of greater than 99% was observed when 40% of the mixture had distilled. Fractional distillation led to a distillate which was 97:3 (A:B).‡This mixture was distilled in the Kugelrohr apparatus at 300 8C and 0.05 mbar for 6–8 h.§This mixture was distilled in the sublimation apparatus at 190–200 8C and #0.001 mbar for 1–3 h.*The undistilled residue contained 56% [C2mim]þ, 44% [NC2 2 2 2]þ, 47% [OTf]2 and 53% [NTf2]2; the distillate contained 30% [C2mim]þ, 70% [NC2 2 2 2]þ, 57% [OTf]2, and 43%[NTf2]2.

Figure 2 | Schematic representation of the differences between protic andaprotic ionic liquids, in both the liquid and the gaseous phases. For theprotic ionic liquids, a dynamic equilibrium exists between the ionic anddissociated forms: [BH]þX2(l) X B(l) þ HX(l) X B(g) þ HX(g). Greencircles represent cations, blue circles represent anions, other coloured circlesrepresent neutral molecules; l, liquid phase; g, gaseous phase. For thegaseous phase over the aprotic ionic liquid, the representation is purelyschematic and has no implication for the actual degree of aggregation.

NATURE|Vol 439|16 February 2006 LETTERS

833

© 2006 Nature Publishing Group

14. Jeapes, A. J. et al. Process for recycling ionic liquids. World Patent WO 01/15175 (2001).

15. Yoshizawa, M., Xu, W. & Angell, C. A. Ionic liquids by proton transfer: Vaporpressure, conductivity, and the relevance of DpKa from aqueous solutions.J. Am. Chem. Soc. 125, 15411–-15419 (2003).

16. Kreher, U. P., Rosamilia, A. E., Raston, C. L., Scott, J. L. & Strauss, C. R.Self-associated, “distillable” ionic media. Molecules 9, 387–-393 (2004).

17. Abdul-Sada, A. K., Elaiwi, A. E., Greenway, A. M. & Seddon, K. R. Evidence forthe clustering of substituted imidazolium salts via hydrogen bonding under theconditions of fast atom bombardment mass spectrometry. Eur. Mass Spectrom.3, 245–-247 (1997).

18. Rebelo, L. P. N., Canongia Lopes, J. N., Esperanca, J. M. S. S. & Filipe, E. On thecritical temperature, normal boiling point, and vapor pressure of ionic liquids.J. Phys. Chem. B 109, 6040–-6043 (2005).

19. Paulechka, Y. U., Zaitsau, Dz. H., Kabo, G. J. & Strechan, A. A. Vapor pressureand thermal stability of ionic liquid 1-butyl-3-methylimidazoliumBis(trifluoromethylsulfonyl)amide. Thermochim. Acta 439, 158–-160 (2005).

20. Maase, M. et al. Method for the separation of acids from chemical reactionmixtures by means of ionic fluids. World Patent WO 03/062171 (2003).

21. Baranyai, K. J., Deacon, G. B., MacFarlane, D. R., Pringle, J. M. & Scott, J. L.

Thermal degradation of ionic liquids at elevated temperatures. Aust. J. Chem.

57, 145–-147 (2004).

Supplementary Information is linked to the online version of the paper at

www.nature.com/nature.

Acknowledgements This work was supported by a grant from FC&T, Portugal

(to J.M.S.S.E. and L.P.N.R.). The Belfast team wishes to thank the industrial

members of QUILL for support and Al Robertson (Cytec) for supplying thephosphonium ionic liquids. For J.W.M. and J.A.W. this work represents an

official contribution of the National Institute of Standards and Technology and is

not subject to copyright in the USA.

Author Information Reprints and permissions information is available at

npg.nature.com/reprintsandpermissions. The authors declare no competing

financial interests. Correspondence and requests for materials should be

addressed to L.P.N.R. (luis.rebelo@itqb.unl.pt).

LETTERS NATURE|Vol 439|16 February 2006

834